def face_compare():
"""比对两张图片是否有人脸和相似度
Args:
url_old: 身份证旧照
url_new: 实时拍摄新照
Return:
比对相似度,json
"""
request_data = request.get_json()
url_old, url_new = '', ''
result = {"code": "400", "status": "failed", "msg": "参数缺少或错误"}
if 'url_old' in request_data:
url_old = request_data['url_old']
if 'url_new' in request_data:
url_new = request_data['url_new']
if url_old is '' or url_new is '':
return jsonify(result)
img_arr_old = face_utils.read_image_from_url(url_old)
face_acc_old = FaceAccredit(img_arr_old)
face_arr_old, face_in_old = face_acc_old.face_encoding()
if (not face_in_old):
result = {
"code": "400",
"result": "failed",
"msg": "no face in url_old"
}
return jsonify(result)
img_arr_new = face_utils.read_image_from_url(url_new)
face_acc_new = FaceAccredit(img_arr_new)
face_arr_new, face_in_new = face_acc_new.face_encoding()
if (not face_in_new):
result = {
"code": "400",
"result": "failed",
"msg": "no face in url_new"
}
return jsonify(result)
dis_result = face_acc_old.face_compare(face_arr_new)
result = {
"code": "0",
"result": "success",
"msg": "compare success",
"prob": str(1 - dis_result[0])
}
return jsonify(result)
def face_binding():
'''Give an image url and idcardid, build the 128-dimension face encoding npy file for face in the image
:param idcard_id 身份证号码
:param face_img_url 单人照片
:return sucess or failed
'''
idcard_id, face_img_url = '', ''
result = {"code": "400", "msg": "参数缺少或错误"}
request_data = request.get_json()
if 'idcard_id' in request_data:
idcard_id = request_data['idcard_id']
if 'face_img_url' in request_data:
face_img_url = request_data['face_img_url']
if idcard_id is '' or face_img_url is '':
return jsonify(result)
img_arr = face_utils.read_image_from_url(face_img_url)
face_acc = FaceAccredit(img_arr)
face_arr_list, face_in_img = face_acc.face_encoding()
if len(face_arr_list) == 1 and face_in_img:
face_utils.save_arr_to_file(face_arr_list[0], idcard_id)
result = {"code": "0", "msg": "绑定成功"}
else:
result = {"code": "500", "msg": "绑定失败"}
return jsonify(result)
def face_livedetect():
'''通过连续抓拍的图片实现活体检测
:param img_urls 视频抓拍照片list, TODO
:param img_url 视频抓拍照片list拼接图片
:param idcard_id (可选)身份证号码, 如果有可绑定,没有不做绑定操作
:return sucess or failed
'''
img_urls, img_url = [], ''
min_faces, threshold = 3, 0.6 # 最小包含人脸, 正确率阈值
result = {"code": "400", "msg": "参数缺少或错误"}
request_data = request.get_json()
if 'img_url' in request_data:
img_url = request_data['img_url']
if img_url is not '' and len(img_url) > 10:
face_acc = FaceAccredit(face_utils.read_image_from_url(img_url))
face_arr_list, face_in_img = face_acc.face_encoding()
if len(face_arr_list) < min_faces:
return jsonify(result)
dis_results = face_acc.face_compare(face_arr_list)
if face_utils.livedetect_result(dis_results, threshold=threshold):
return jsonify({
"code": "0",
"result": True,
"data": dis_results.tolist()
})
return jsonify({
"code": "0",
"result": False,
"data": dis_results.tolist()
})
if 'img_urls' in request_data:
img_urls = request_data['img_urls']
if type(img_urls) is not list or len(img_urls) < min_faces:
return jsonify(result)
face_acc, face_arr_list, is_ok, err_url = face_arr_build(img_urls)
if is_ok is False:
result['errurl'] = err_url
return jsonify(result)
dis_results = face_acc.face_compare(face_arr_list)
if face_utils.livedetect_result(dis_results, threshold=threshold):
return jsonify({
"code": "0",
"result": True,
"data": dis_results.tolist()
})
return jsonify({
"code": "0",
"result": False,
"data": dis_results.tolist()
})
def face_arr_build(face_urls):
'''根据批量人脸图片创建认证数组
param: face_urls ["http://xxx/xx/face00.jpg", "https://xxx/xx/face01.jpg"]
'''
face_arrs, face_acc = [], None
for face_url in face_urls:
face_acc = FaceAccredit(face_utils.read_image_from_url(face_url))
face_arr_list, face_in_img = face_acc.face_encoding()
if face_in_img and len(face_arr_list) == 1:
face_arrs.append(face_arr_list[0])
else:
return face_acc, face_arrs, False, face_url
return face_acc, face_arrs, True, None
def face_identification():
'''根据已绑定的`128-dimension face encoding npy file`查询已存在的人脸信息
:param face_img_url 人脸照片
:return 相似度大于指定阈值的身份证编码 eg: [{"idcard_id":"450225198208107439","prob": 0.5867}]
'''
face_img_url = ''
result = {"code": "400", "msg": "参数缺少或错误"}
request_data = request.get_json()
if 'face_img_url' in request_data:
face_img_url = request_data['face_img_url']
if face_img_url is '':
return jsonify(result)
img_arr = face_utils.read_image_from_url(face_img_url)
face_acc = FaceAccredit(img_arr)
face_arr_list, face_in_img = face_acc.face_encoding()
if face_in_img:
face_npy_list, face_name_list = face_utils.get_npy_list()
dis_results = face_acc.face_compare(face_npy_list)
if dis_results.ndim == 1:
identifactions = face_utils.identifaction_result_build(
dis_results.tolist(), face_name_list, 0.55)
result = {
"code": "0",
"msg": "identify success",
"data": identifactions
}
else:
result = {
"code": "500",
"msg":
"identify failed, result ndim is " + str(dis_results.ndim)
}
else:
result = {"code": "400", "msg": "no face in face_img_url"}
return jsonify(result)